
import os
import pandas as pd
%matplotlib inline
import pandas as pd
import numpy as np
import math
import scipy
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import plotly.graph_objects as go
import plotly.io as pio
from plotly.subplots import make_subplots
from IPython.display import display

# Packages for machine learning method k-means clustering
from sklearn import preprocessing
from sklearn.metrics import classification_report
from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_score, davies_bouldin_score
from sklearn.mixture import GaussianMixture


file_path = os.getcwd() + '\Data.csv'

# Read data: remove first 6 rows and last row that do not contain data
data = pd.read_csv(file_path, delimiter = ';', skiprows = 6, skipfooter = 1, engine = 'python')


# Make person characteristics index
data.set_index('Vervoerwijzen', inplace = True)

# Remove rows with missing information
data_updated = data.drop(labels = ["Persoonskenmerken", "Totaal personen", "Leeftijd:6 tot 12 jaar", "Leeftijd: 12 tot 18 jaar",
                                   "Participatie: werkloos", "Participatie:  arbeidsongeschikt", 
                                   "Geen rijbewijs; jonger dan 17 jaar", "OV-Studentenkaart: weekendabonnement"])

# Replace dots in dataset with 0
data_updated = data_updated.replace('.', 0)

# Only keep columns with average number of movements per day per person
columns_to_keep = [col for col in data_updated.columns if 
                   data_updated.loc["Onderwerp", col] == "Gemiddeld per persoon per dag |Verplaatsingen "]
data_updated = data_updated[columns_to_keep]

# Remove the "onderwerp" row, since this is now the same for all columns
data_updated = data_updated.drop("Onderwerp")

# The data is formatted as strings, we convert these to floats and make NaN's equal to 0 for all three dataframes
def preprocess_dataframe(df):
    # Convert strings to floats and replace NaN values with 0
    df = df.apply(lambda col: col.str.replace(',', '.').astype(float).fillna(0))
    return df

data_updated = preprocess_dataframe(data_updated)

# Merge the rows "Leeftijd: 65 tot 75 jaar" en "75jaar of ouder" into one row
data_updated = (data_updated.reset_index()
                .replace({"Vervoerwijzen": {"Leeftijd: 75 jaar of ouder" : "Leeftijd: 65 tot 75 jaar"}})
                .groupby("Vervoerwijzen", sort=False).sum()
)
data_updated = data_updated.rename({"Leeftijd: 65 tot 75 jaar" : "Leeftijd: 65 jaar of ouder"})

# Split dataframe into 5 separate datframes: one for each year and remove row "perioden"
columns_2018 = [col for col in data_updated.columns if int(data_updated.loc["Perioden", col]) == 2018]
columns_2019 = [col for col in data_updated.columns if int(data_updated.loc["Perioden", col]) == 2019]
columns_2020 = [col for col in data_updated.columns if int(data_updated.loc["Perioden", col]) == 2020]
columns_2021 = [col for col in data_updated.columns if int(data_updated.loc["Perioden", col]) == 2021]
columns_2022 = [col for col in data_updated.columns if int(data_updated.loc["Perioden", col]) == 2022]
transport_2018 = data_updated[columns_2018].drop("Perioden")
transport_2019 = data_updated[columns_2019].drop("Perioden")
transport_2020 = data_updated[columns_2020].drop("Perioden")
transport_2021 = data_updated[columns_2021].drop("Perioden")
transport_2022 = data_updated[columns_2022].drop("Perioden")

# Merge car driver and passenger into car, cycle and walk into active modes and trainand bus/tram/metro into public transport
def merge_transport_categories(df):
    # The different years have a number behind the column names. Remove these numbers.
    column_names = ["Total", "Personenauto (bestuurder)", "Personenauto  (passagier)", 
                    "Trein", "Bus/tram/metro", "Fiets", "Lopen", "Overige vervoerwijze"]
    df.columns = column_names
    # Merge desired columns
    df["Passengercar"] = df["Personenauto (bestuurder)"] + df["Personenauto  (passagier)"]
    df["Public Transport"] = df["Trein"] + df["Bus/tram/metro"]
    df["Active modes"] = df["Fiets"] + df["Lopen"]
    df = df.drop(columns = ["Personenauto (bestuurder)", "Personenauto  (passagier)", "Trein", 
                            "Bus/tram/metro", "Fiets", "Lopen"])
    # Move "Overige vervoerswijze" to the end of the dataframe
    Other = df.pop("Overige vervoerwijze")
    df["Other modes"] = Other
    # Translate all index names to English
    index_names = ["Age: 18 to 25 years", "Age: 25 to 35 years", "Age: 35 to 50 years", "Age: 50 to 65 years", 
                   "Age: 65 years or older", "Migrationbackground: The Netherlands", "Migrationbackground: Western", 
                   "Migrationbackground: Non-western", "Standardised income: 1st 20% group", 
                   "Standardised income: 2nd 20% group","Standardised income: 3rd 20% group", 
                   "Standardised income: 4th 20% group", "Standardised income: 5th 20% group", 
                   "PT-studentcard: Weekday subscription", "PT-studentcard: None","Education level: Low", 
                   "Education level: middle", "Education level: High", "Participation: Working 12 to 30 h/w",
                   "Participation: Working 30+ h/w", "Participation: student", "Participation: retired",
                   "Participation: Other", "Drivers licence; owns own car", "Drivers license; car in household",
                   "Drivers license; no car", "No drivers license"]
    df.index = index_names

    return df

transport_2018 = merge_transport_categories(transport_2018)
transport_2019 = merge_transport_categories(transport_2019)
transport_2020 = merge_transport_categories(transport_2020)
transport_2021 = merge_transport_categories(transport_2021)
transport_2022 = merge_transport_categories(transport_2022)


# Example of what a dataset now looks like. Layout the same for all years, only numbers might differ.
transport_2018


# Normalise dataframe rows
def normal_rows(df):
    normal_df = df.drop(columns = ["Total"])
    normal_df = normal_df.div(normal_df.sum(axis = 1), axis = 0)
    return normal_df

normal_2018 = normal_rows(transport_2018)
normal_2019 = normal_rows(transport_2019)
normal_2020 = normal_rows(transport_2020)
normal_2021 = normal_rows(transport_2021)
normal_2022 = normal_rows(transport_2022)

normal_2018


# Filtered DataFrame for each year
dfs = [normal_2018, normal_2019, normal_2020, normal_2021, normal_2022]

# Changing layout of plots for better readability 
fig, axes = plt.subplots(3, 2, figsize=(20, 20)) # Two rows to show the 5 plots to prevent cluttering
fig.subplots_adjust(hspace=1.1)  
axes = axes.flatten()

for i, ax in enumerate(axes):
    if i < len(dfs):
        df = dfs[i]
        df.plot(kind='bar', stacked=True, color = ["salmon", "gold", "skyblue", "lawngreen"], fontsize = 10, ax=ax)
        ax.set_title(f"Mode choice distribution {2018 + i}")
        ax.set_xlabel("Person characteristics")
        ax.set_ylabel("Choice distribution")
        ax.legend(loc='lower left')
        ax.tick_params(axis='x', rotation=90)
    else:
        fig.delaxes(ax)  

# Remove the large plots 
plt.close('all')

# Display the combined grid of small plots
display(fig)


# Define the list of years
years = [2018, 2019, 2020, 2021, 2022]

# Define the list of age groups to keep
age_groups_to_keep = [
    "Age: 18 to 25 years",
    "Age: 25 to 35 years",
    "Age: 35 to 50 years",
    "Age: 50 to 65 years",
    "Age: 65 years or older"
]

# Define the desired colors
custom_colors = ["salmon", "gold", "skyblue", "lawngreen"]

# Create a separate plot for each year
for year in years:
    if year == 2018:
        year_df = transport_2018
    elif year == 2019:
        year_df = transport_2019
    elif year == 2020:
        year_df = transport_2020
    elif year == 2021:
        year_df = transport_2021
    else:
        year_df = transport_2022

    # Remove the "Total" column
    year_df = year_df.drop("Total", axis=1)

    # Calculate the percentage of transportation mode used by each age group
    total_per_age_group = year_df.sum(axis=1)
    percentage_df = year_df.divide(total_per_age_group, axis=0) * 100

    # Keep only the specified age groups
    percentage_df = percentage_df.loc[age_groups_to_keep]

    # Create an interactive bar chart for the current year
    fig = px.bar(
        percentage_df,
        x=percentage_df.index,  # Use the age groups as x-axis categories
        y=percentage_df.columns,
        labels={"value": "Percentage", "index": "Age Group"},
        color_discrete_sequence=custom_colors,  # Use custom colors
        title=f"Modes of Transportation by Age Group ({year})"
    )

    # Show the interactive plot
    fig.show()


# Filter Income Groups
income_groups = [
    "Standardised income: 1st 20% group",
    "Standardised income: 2nd 20% group",
    "Standardised income: 3rd 20% group",
    "Standardised income: 4th 20% group",
    "Standardised income: 5th 20% group"
]

# Filtered DataFrame for each year
dfs = [transport_2018, transport_2019, transport_2020, transport_2021, transport_2022]

# Changing layout of plots for better readability 
fig, axes = plt.subplots(3, 2, figsize=(20, 20)) # Two rows to show the 5 plots to prevent cluttering
fig.subplots_adjust(hspace=1.1)  
axes = axes.flatten()

for i, ax in enumerate(axes):
    if i < len(dfs):
        df = dfs[i]
        transport_filtered = df.loc[income_groups]
        transport_filtered.plot(kind='bar', stacked=False, color=["purple", "salmon", "gold", "skyblue", "lawngreen"], ax=ax)
        ax.set_title(f"Income Groups & Mode of Transport {2018 + i}")
        ax.set_xlabel("Income Groups")
        ax.set_ylabel("Average number of movements per day")
        ax.legend(title='Transport Modes', loc='upper right')
        ax.tick_params(axis='x', rotation=45)
    else:
        fig.delaxes(ax)  

# Remove the large plots 
plt.close('all')

# Display the combined grid of small plots
display(fig)


# Data for 'PT-studentcard: Weekday subscription,' 'Drivers licence; owns own car,' 'No drivers license,' and 'Drivers license; no car' for each year
years = ['2018', '2019', '2020', '2021', '2022']

# Extract the "Total" row for each category
pt_student_card = [transport_2018.loc['PT-studentcard: Weekday subscription', 'Total'],
                   transport_2019.loc['PT-studentcard: Weekday subscription', 'Total'],
                   transport_2020.loc['PT-studentcard: Weekday subscription', 'Total'],
                   transport_2021.loc['PT-studentcard: Weekday subscription', 'Total'],
                   transport_2022.loc['PT-studentcard: Weekday subscription', 'Total']]

owning_car = [transport_2018.loc['Drivers licence; owns own car', 'Total'],
             transport_2019.loc['Drivers licence; owns own car', 'Total'],
             transport_2020.loc['Drivers licence; owns own car', 'Total'],
             transport_2021.loc['Drivers licence; owns own car', 'Total'],
             transport_2022.loc['Drivers licence; owns own car', 'Total']]

# Create a DataFrame for the data
data = pd.DataFrame({'Year': years,
                     'PT Student Card': pt_student_card,
                     'Owns a Car': owning_car})


# Create an interactive line chart using Plotly Express with custom colors
fig = px.line(data, x='Year', y=data.columns[1:], labels={'value': 'Total Count'},
              title='PT Student Card vs Car Ownership over the Years',
              color_discrete_map={'PT Student Card': 'gold', 'Owns a Car': 'salmon'})
fig.update_traces(mode='markers+lines', marker=dict(size=10))
fig.show()


# Select the rows for the specified categories
categories = [
    "PT-studentcard: Weekday subscription",
    "Drivers licence; owns own car"]

# Select the relevant data for the specified categories, excluding the "Total" column
data_to_plot = transport_2022.loc[categories].drop(columns=["Total"])

# Transpose the data
data_to_plot = data_to_plot.T

# Calculate the percentage distribution for each mode
total = data_to_plot.sum(axis=1)
data_to_plot_percentage = data_to_plot.div(total, axis=0) * 100

# Define the colors
colors = ['gold','salmon']

# Create the stacked bar plots with modes as stacked bars
data_to_plot_percentage.plot(kind="bar", stacked=True, figsize=(10, 6), color=colors)

# Customize the plot
plt.title("People's preference on Modes of Transportation based on PT-student cards and cars (2018)")
plt.xlabel("Modes")
plt.ylabel("Percentage Distribution")
plt.legend(title="Characteristics")

# Show the plot
plt.show()


# Filter the three different Migration backgrounds
migration_background = [
    "Migrationbackground: The Netherlands",
    "Migrationbackground: Western",
    "Migrationbackground: Non-western",
]

# Create a grid for the three line plots
fig, axes = plt.subplots(2, 2, figsize=(18, 17))
fig.delaxes(axes[1, 1])  # Remove the fourth empty subplot
colors = ["salmon", "gold", "skyblue", "lawngreen", "purple"]

for i, mbg in enumerate(migration_background):
    ax = axes[i // 2, i % 2]
    ax.set_title(f" {mbg}")
    ax.set_xlabel("Transport Modes")
    ax.set_ylabel("Average number of movements for each year")
    
    for j, df in enumerate(dfs):
        transport_filtered = df.loc[mbg]
        transport_filtered = transport_filtered.T 
        transport_filtered = transport_filtered.drop("Total", errors="ignore")
        transport_filtered.plot(kind='line', color=colors[j], ax=ax, marker='o', label=f"Year {2018 + j}")

    # Create a legend for each subplot
    ax.legend(loc='upper right')

# Hide the individual large plots
plt.close('all')

# Display the combined grid of small plots
display(fig)


## Creating first plot
# Creating one dataframe containing the total number of transport movements per mode per year
combined_df = pd.DataFrame()
i = 2018
for frame in [transport_2018, transport_2019, transport_2020, transport_2021, transport_2022]: # Iterate over the different years
    data = [sum(frame["Passengercar"]), sum(frame["Public Transport"]), sum(frame["Active modes"]), 
            sum(frame["Other modes"])] # Sum the movements per mode
    combined_df[str(i)] = data # Add to dataframe
    i += 1

index_names = ["Passengercar", "Public Transport", "Active modes", "Other modes"]
combined_df.index = index_names # Give the index the correct names

## Creating second plot
# The see the distribution over the modes, the data needs to be normalised per row
def normal_rows_years(df):
    normal_df = df.div(df.sum(axis = 1), axis = 0) # Normalising each row
    return normal_df

normal_combined_df = normal_rows_years(combined_df.T) # Normalising the combined dataframe

# Plotting both plots next to each other
fig, axes = plt.subplots(1, 2, figsize=(15, 5))  # Create a subplot with two axes

combined_df.T.plot(kind="bar", stacked=True, color=["salmon", "gold", "skyblue", "lawngreen"],
                   ax=axes[0], fontsize=10)  # Plot the first stacked bar plot on the left axis
axes[0].set_title("Travel behavior over the years")
axes[0].set_xlabel("Years")
axes[0].set_ylabel("Number of travel movements")

normal_combined_df.plot(kind="bar", stacked=True, color=["salmon", "gold", "skyblue", "lawngreen"],
                        ax=axes[1], fontsize=10)  # Plot the second stacked bar plot on the right axis
axes[1].set_title("Normalized travel behavior over the years")
axes[1].set_xlabel("Years")
axes[1].set_ylabel("Number of travel movements")

plt.tight_layout()
plt.show()


# Correlation plot of mode choices
fig, axes = plt.subplots(1)
corr_2018 = normal_2018.corr()
heatmap_2018 = sns.heatmap(round(corr_2018, 2), annot=True, cmap='Blues', ax=axes, fmt='.2f')


# Elbow method to determine number of clusters
wcss = []
N_max = 10
for i in range(1, N_max):
    kmeans = KMeans(n_clusters=i,random_state=1).fit(normal_2018)
    wcss.append(kmeans.inertia_)
    
plt.figure()
plt.plot(range(1, N_max),wcss)
plt.title('The Elbow Method')
plt.xlabel('Number of clusters')
plt.ylabel('J')
plt.show()

# The warning that is obtained is researched: whatever the memory leak is, it's tiny.
# For short-running jobs, the script will probably end before the leak gets too bad, and the process will free the memory.

C:\Users\Gebruiker\anaconda3\lib\site-packages\sklearn\cluster\_kmeans.py:1036: UserWarning:

KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.


# Calculate different types of scores to decide on number of clusters
def get_km(k, X): # Function to fit KMeans
    km = KMeans(n_clusters=k, random_state=1)
    km.fit(X)
    return km

def get_bic(k, X): # Function to calculate BIC
    gmm = GaussianMixture(n_components=k, init_params='kmeans', random_state=1)
    gmm.fit(X)
    return gmm.bic(X)

def get_score(k, X): # Function to get other scores (Davies, Silhouette)
    km = get_km(k, X)
    y_pred = km.predict(X)
    bic = get_bic(k, X)
    sil = silhouette_score(X, y_pred)
    db = davies_bouldin_score(X, y_pred)
    return k, bic, sil, db

# Create dataframe with all scores for different values of k
scores_df = pd.DataFrame([get_score(k, normal_2018) for k in range(2, 11)],
                  columns=['k', 'BIC', 'silhouette','davies'])

def plot_compare(df, y1, y2, x, fig, ax1): # Function that compares scores with scaled axes
    ax1.plot(df[x], df[y1], color='tab:red')
    ax1.set_title(f'{y1} and {y2}')
    ax1.set_xlabel(x)
    ax1.set_ylabel(y1, color='tab:red')
    ax1.tick_params(axis='y', labelcolor='tab:red')

    ax2 = ax1.twinx()
    ax2.plot(df[x], df[y2], color='tab:blue')
    ax2.set_ylabel(y2, color='tab:blue')
    ax2.tick_params(axis='y', labelcolor='tab:blue')

def plot_result(df, y1, y2, y3, x): # Function to plot graphs side by side
    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))  # Create a subplot with two axes

    plot_compare(df, y1, y2, x, fig, ax1)  # Plot the first graph on the first axis
    plot_compare(df, y1, y3, x, fig, ax2)  # Plot the second graph on the second axis

    plt.tight_layout()
    plt.show()

plot_result(scores_df, 'BIC', 'silhouette', 'davies', 'k')


# Apply k-means clustering technique
kmeans = KMeans(n_clusters=4, random_state=1).fit(normal_2018)

# Obtain clusters and add label to data
normal_2018['cluster'] = kmeans.labels_

# Make binary columns in data for each cluster 
normal_2018['Cluster0'] = np.where(normal_2018['cluster'] != 0, 0, 1)
normal_2018['Cluster1'] = np.where(normal_2018['cluster'] != 1, 0, 1)
normal_2018['Cluster2'] = np.where(normal_2018['cluster'] != 2, 0, 1)
normal_2018['Cluster3'] = np.where(normal_2018['cluster'] != 3, 0, 1)

# Make cluster vs. mode correlation plot to see which modes are most present in each cluster
fig, axes = plt.subplots(figsize = (5,4))
correlation_table = normal_2018.corr()
corr2 = correlation_table.drop(correlation_table.index[range(5)])
corr3 = corr2.drop(columns=corr2.columns[4:])
corr_m = sns.heatmap(round(corr3, 2), annot=True, cmap='Blues', ax=axes, fmt='.2f')


# Make dataframe for every cluster including personal characteristics corresponding to cluster
cluster0 = normal_2018.loc[normal_2018['cluster'] == 0].drop(columns=normal_2018.columns[4:])
cluster1 = normal_2018.loc[normal_2018['cluster'] == 1].drop(columns=normal_2018.columns[4:])
cluster2 = normal_2018.loc[normal_2018['cluster'] == 2].drop(columns=normal_2018.columns[4:])
cluster3 = normal_2018.loc[normal_2018['cluster'] == 3].drop(columns=normal_2018.columns[4:])


# Make functions for analysis

# Number of characteristics in cluster
def length(cluster):
    output = 'The number of characteristics that are in this cluster is equal to %i from the %i characteristics in the dataset.' \
            % (len(cluster), len(normal_2018))
    print(output)
    
# Average normalized distribution of movements over the four different modes in cluster
def distribution(cluster):
    distribution = {}
    for i in cluster:
        distribution[str(i)] = cluster[i].mean()
    return distribution
    
# Function that returns top two modes with most movements in cluster
def mode_choice(cluster):
    dist = distribution(cluster)
    sort = sorted(((v,k) for k,v in dist.items()))
    mode1 = max(dist, key=dist.get)
    mode2 = sort[-2][1]
    return mode1, mode2

# Print characteristics (index) in cluster
def characteristics(cluster):
    characteristics = cluster.index.values.tolist()
    return characteristics

# Three most influential characteristics for mode choice in cluster
def significant_characteristics(cluster):
    lst = []
    chars = dict(cluster[str(mode_choice(cluster)[0])].nlargest(3)).keys()
    for key in characteristics:
        lst.append(key)
    print('The three characteristics with highest movements for mode "%s" are "%s", "%s" and "%s".' 
          % (str(mode_choice(cluster)[0]), lst[0], lst[1], lst[2]))


length(cluster0)
length(cluster1)
length(cluster2)
length(cluster3)

The number of characteristics that are in this cluster is equal to 12 from the 27 characteristics in the dataset.
The number of characteristics that are in this cluster is equal to 4 from the 27 characteristics in the dataset.
The number of characteristics that are in this cluster is equal to 3 from the 27 characteristics in the dataset.
The number of characteristics that are in this cluster is equal to 8 from the 27 characteristics in the dataset.


print(characteristics(cluster0))
print('The mode choice of cluster 0 is mostly %s, in second place is %s.' 
      % (mode_choice(cluster0)[0], mode_choice(cluster0)[1]))

print(characteristics(cluster1))
print('The mode choice of cluster 1 is mostly %s, in second place is %s.' 
      % (mode_choice(cluster1)[0], mode_choice(cluster1)[1]))

print(characteristics(cluster2))
print('The mode choice of cluster 2 is mostly %s, in second place is %s.' 
      % (mode_choice(cluster2)[0], mode_choice(cluster2)[1]))

print(characteristics(cluster3))
print('The mode choice of cluster 3 is mostly %s, in second place is %s.' 
      % (mode_choice(cluster3)[0], mode_choice(cluster3)[1]))

['Age: 65 years or older', 'Migrationbackground: The Netherlands', 'Migrationbackground: Western', 'Standardised income: 2nd 20% group', 'Standardised income: 3rd 20% group', 'Standardised income: 4th 20% group', 'PT-studentcard: None', 'Education level: Low', 'Education level: High', 'Participation: Working 12 to 30 h/w', 'Participation: retired', 'Participation: Other']
The mode choice of cluster 0 is mostly Passengercar, in second place is Active modes.
['Standardised income: 1st 20% group', 'Participation: student', 'Drivers license; no car', 'No drivers license']
The mode choice of cluster 1 is mostly Active modes, in second place is Passengercar.
['Age: 18 to 25 years', 'Migrationbackground: Non-western', 'PT-studentcard: Weekday subscription']
The mode choice of cluster 2 is mostly Active modes, in second place is Passengercar.
['Age: 25 to 35 years', 'Age: 35 to 50 years', 'Age: 50 to 65 years', 'Standardised income: 5th 20% group', 'Education level: middle', 'Participation: Working 30+ h/w', 'Drivers licence; owns own car', 'Drivers license; car in household']
The mode choice of cluster 3 is mostly Passengercar, in second place is Active modes.


# Make interactive barplot using distribution data of different clusters
data = {
    'Cluster': ['Cluster0', 'Cluster1', 'Cluster2', 'Cluster3'],
    'Passengercar': [distribution(cluster0)['Passengercar'], distribution(cluster1)['Passengercar'], 
                     distribution(cluster2)['Passengercar'], distribution(cluster3)['Passengercar']],
    'Public Transport': [distribution(cluster0)['Public Transport'], distribution(cluster1)['Public Transport'], 
                         distribution(cluster2)['Public Transport'], distribution(cluster3)['Public Transport']],
    'Active Modes': [distribution(cluster0)['Active modes'], distribution(cluster1)['Active modes'], 
                     distribution(cluster2)['Active modes'], distribution(cluster3)['Active modes']],
    'Other Modes': [distribution(cluster0)['Other modes'], distribution(cluster1)['Other modes'], 
                    distribution(cluster2)['Other modes'], distribution(cluster3)['Other modes']]
}

df = pd.DataFrame(data)
colors = ["salmon", "gold", "skyblue", "lawngreen"]
df = pd.melt(df, id_vars=['Cluster'], var_name='Mode', value_name='Distribution')

# Create an interactive barplot using Plotly Express
fig = px.bar(df, x='Cluster', y='Distribution', color='Mode',
             color_discrete_sequence=colors,
             title='Movements by Mode in Each Cluster',
             labels={'Cluster': 'Cluster', 'Distribution': 'Distribution'},
             category_orders={'Cluster': ['Cluster0', 'Cluster1', 'Cluster2', 'Cluster3']}
            )
fig.show()

	Total	Passengercar	Public Transport	Active modes	Other modes
Age: 18 to 25 years	2.74	1.05	0.44	1.12	0.13
Age: 25 to 35 years	2.99	1.52	0.24	1.10	0.12
Age: 35 to 50 years	3.20	1.71	0.13	1.22	0.14
Age: 50 to 65 years	2.80	1.45	0.11	1.12	0.12
Age: 65 years or older	4.05	1.85	0.12	1.94	0.13
Migrationbackground: The Netherlands	2.90	1.38	0.12	1.28	0.12
Migrationbackground: Western	2.57	1.09	0.20	1.21	0.08
Migrationbackground: Non-western	2.22	0.82	0.31	1.03	0.07
Standardised income: 1st 20% group	2.45	0.79	0.22	1.33	0.10
Standardised income: 2nd 20% group	2.46	1.07	0.14	1.16	0.10
Standardised income: 3rd 20% group	2.88	1.34	0.14	1.27	0.12
Standardised income: 4th 20% group	3.01	1.49	0.14	1.27	0.12
Standardised income: 5th 20% group	3.00	1.57	0.15	1.19	0.10
PT-studentcard: Weekday subscription	2.76	0.80	0.80	1.07	0.09
PT-studentcard: None	2.78	1.30	0.12	1.24	0.11
Education level: Low	2.24	0.94	0.11	1.04	0.14
Education level: middle	2.91	1.49	0.18	1.10	0.15
Education level: High	3.10	1.56	0.20	1.29	0.06
Participation: Working 12 to 30 h/w	3.35	1.69	0.14	1.44	0.07
Participation: Working 30+ h/w	2.98	1.68	0.17	0.98	0.16
Participation: student	2.85	0.76	0.26	1.76	0.08
Participation: retired	2.14	1.00	0.06	1.02	0.05
Participation: Other	2.52	1.12	0.09	1.21	0.10
Drivers licence; owns own car	2.90	1.79	0.06	0.93	0.11
Drivers license; car in household	2.99	1.59	0.14	1.15	0.12
Drivers license; no car	2.84	0.57	0.40	1.67	0.21
No drivers license	2.13	0.43	0.32	1.28	0.09

	Passengercar	Public Transport	Active modes	Other modes
Age: 18 to 25 years	0.383212	0.160584	0.408759	0.047445
Age: 25 to 35 years	0.510067	0.080537	0.369128	0.040268
Age: 35 to 50 years	0.534375	0.040625	0.381250	0.043750
Age: 50 to 65 years	0.517857	0.039286	0.400000	0.042857
Age: 65 years or older	0.457921	0.029703	0.480198	0.032178
Migrationbackground: The Netherlands	0.475862	0.041379	0.441379	0.041379
Migrationbackground: Western	0.422481	0.077519	0.468992	0.031008
Migrationbackground: Non-western	0.367713	0.139013	0.461883	0.031390
Standardised income: 1st 20% group	0.323770	0.090164	0.545082	0.040984
Standardised income: 2nd 20% group	0.433198	0.056680	0.469636	0.040486
Standardised income: 3rd 20% group	0.466899	0.048780	0.442509	0.041812
Standardised income: 4th 20% group	0.493377	0.046358	0.420530	0.039735
Standardised income: 5th 20% group	0.521595	0.049834	0.395349	0.033223
PT-studentcard: Weekday subscription	0.289855	0.289855	0.387681	0.032609
PT-studentcard: None	0.469314	0.043321	0.447653	0.039711
Education level: Low	0.421525	0.049327	0.466368	0.062780
Education level: middle	0.510274	0.061644	0.376712	0.051370
Education level: High	0.501608	0.064309	0.414791	0.019293
Participation: Working 12 to 30 h/w	0.505988	0.041916	0.431138	0.020958
Participation: Working 30+ h/w	0.561873	0.056856	0.327759	0.053512
Participation: student	0.265734	0.090909	0.615385	0.027972
Participation: retired	0.469484	0.028169	0.478873	0.023474
Participation: Other	0.444444	0.035714	0.480159	0.039683
Drivers licence; owns own car	0.619377	0.020761	0.321799	0.038062
Drivers license; car in household	0.530000	0.046667	0.383333	0.040000
Drivers license; no car	0.200000	0.140351	0.585965	0.073684
No drivers license	0.202830	0.150943	0.603774	0.042453

Predicting mode choice behavior based on person characteristics¶

Research Objective¶

Contribution Statement¶

Data Used¶

Data Pipeline¶

Import packages¶

Import data¶

Data processing¶

Data analysis¶

Subquestion 1: What is the probability for a person to choose a certain transport mode based on one personal characteristic?¶

Subquestion 2: From the previous subquestion, which personal characteristics jump out and could be researched further?¶

Age groups¶

Income groups¶

Public transport vs. car ownership¶

People's preference on Modes of Transportation¶

Migration background¶

Subquestion 3: Is there a significant difference over the years in mode choice?¶

Subquestion 4: Can the groups obtained through K-means clustering based on personal characteristics effectively reflect distinct transportation mode preferences?¶

K-means clustering¶

Conlusion¶

Limitations and recommendations¶

References¶